ObjectivesRheumatoid arthritis (RA) and thyroid dysfunction are frequently observed in the same patient. However, whether they co‐occur or exhibit a causal relationship remains uncertain. We aimed to systematically investigate the causal relationship between RA and thyroid function using a large sample and advanced methods.MethodsBi‐directional two‐sample Mendelian randomization (MR) analysis was performed based on RA and six thyroid function traits datasets in the European population. The robustness of the results was demonstrated using multiple MR methods and a series of sensitivity analyses. Multivariable MR using Bayesian model averaging (MR‐BMA) was performed to adjust for possible competing risk factors. A sensitivity dataset, which included patients with seropositive RA and controls, was used to repeat the analyses. Furthermore, enrichment analysis was employed to discover the underlying mechanism between RA and thyroid functions.ResultsA significantly positive causal effect was identified for RA on autoimmune thyroid disease (AITD), as well as for AITD on RA (P < 0.001). Further sensitivity analyses showed consistent causal estimates from a variety of MR methods. After removing the outliers, MR‐MBA results showed that RA and AITD were independent risk factors in their bi‐directional causality, even in the presence of other competing risk factors (Padj < 0.05). Enrichment analysis showed immune cell activation and immune response play crucial roles in them.ConclusionOur results illustrate the significant bi‐directional causal effect of RA and AITD, which holds even in multiple competing risk factors. Clinical screening for thyroid dysfunction in RA patients deserves further attention, and vice versa.